Ok first off sorry if i accidentally make a double post, my internet connection is unstable.
I am working on a stereo LM3886T stereo amp (schematic below although i think everyone knows it, seems to be popular IC) but not in the classic and easy i think dual supply mode. I am limited by the power supply itself since using my biggest transformer would only deliver 2x18V (before apply signal).
1. i know signal ground must be separated from power ground or else because interaction sound quality would be mediocre ( i already experienced that) but which components? I do believe everything except Zobel and mute cap is that correct?
2. if i use a twin shielded cable for input do i connect each shield to each signal ground via the small resistor that must be used to separate power and signal paths? or do i connect them straight to signal grounds and each "separation" resistor goes to the power gnd of it's IC?
in the first case i believe it could possibly be beneficial (and i've seen some local enthusiasts do that but with dual supply circuit) since the path will still end in the ground but that of the preamp which is tube so there is very little current flowing there so it is quiet.
3. the mute resistor? if i judge by the datasheet it should be about 5.6K (keep in mind the voltage at pin 4 is 0) but i used 4.7K and seems to work just fine.
I am working on a stereo LM3886T stereo amp (schematic below although i think everyone knows it, seems to be popular IC) but not in the classic and easy i think dual supply mode. I am limited by the power supply itself since using my biggest transformer would only deliver 2x18V (before apply signal).
An externally hosted image should be here but it was not working when we last tested it.
1. i know signal ground must be separated from power ground or else because interaction sound quality would be mediocre ( i already experienced that) but which components? I do believe everything except Zobel and mute cap is that correct?
2. if i use a twin shielded cable for input do i connect each shield to each signal ground via the small resistor that must be used to separate power and signal paths? or do i connect them straight to signal grounds and each "separation" resistor goes to the power gnd of it's IC?
in the first case i believe it could possibly be beneficial (and i've seen some local enthusiasts do that but with dual supply circuit) since the path will still end in the ground but that of the preamp which is tube so there is very little current flowing there so it is quiet.
3. the mute resistor? if i judge by the datasheet it should be about 5.6K (keep in mind the voltage at pin 4 is 0) but i used 4.7K and seems to work just fine.
The three most important points to single point ground...the bottom of the input pot, the bottom of Ci, and the grounded side of the speaker.
Just behind that, RA and CA should be there...
The biggest issue is managing the large half wave rectified currents from the negative supply pin. You need to keep those out of the signal path.
Just behind that, RA and CA should be there...
The biggest issue is managing the large half wave rectified currents from the negative supply pin. You need to keep those out of the signal path.
Last edited:
Not sure i follow. From what you are describing it sounds like star grounding (?). The issue is these are 3886 T not TF, that means the power ground is in contact with the heatsink. That brings the problem of a possible ground loop between the tabs and the input ground. That is why perhaps i should rephrase: am i in this setup allowed to break ground loop by a small resistor as i would in the dual supply setup? The fact that there is no audible hum does not mean there is no current induced and as the psu is switching, the "hum" is not audible but may still superimpose on the signal and impair (especially the highs)
That means that you *do* have two rails.
Go for the simpler split supply configuration.
Which will solve 2 problems:
1) your back flange will be insulated from ground.
2) more important: by definition, out of the 3 high current paths you have there, 2 (+/- rails) will *not* be mixed with ground and the 3rd, the speaker ground or return, can use its own wire straight to the filter caps centerpoint.
Remaining ground is audio ground, not a difficult task since currents are very low, simplifying your task.
Perhaps i should, although that wouldn't provide the speakers with half of what they can take. Then there's the problem of preamp power. Since the transformer mentioned is large, the switching supply won't fit any more so then a second transformer is needed for preamp. And now none of the supplies would be regulated. To make matters worse, the em fields of transformers interact and cause the case to hum, or more accurately the entire desk hums, extremely unpleasant.
A 18+18Vac transformer will allow the same maximum power output as a 36Vac transformer.
The single secondary version uses the single polarity schematic as you posted.
The dual secondary (or a centre tapped secondary) can use the same single polarity schematic, or the "normal" dual polarity schematic.
a 36Vac transformer would result in a voltage of up to 40 after rectification/filtering. the output from my current power supply is 60V regulated so there is a little bit of a difference.
as for my transformer, the 18+18V is not the ac output but what is obtained after rectification, so it will drop even lower when there is signal. the transformer actually delivers 14+14Vac. that is not the main problem since i don't plan on going to insane levels (my room is rather small) but then i have to add the transformer for the preamp as well and it's not even the same 180V as i use now, but 300V so i have to redo the whole preamp as well and that's not particularly nice.
as a solution i will start rummaging around for any piece of mica i can use to insulate the ic's from heatsink so i'll just need to insulate the shields of the input cable from the ground by resistors as mentioned and the preamp will receive ground for power via a separate wire from the power supply. that should take care of any possible ground loop. As for the grounds in the output stages, i suppose i don't necessarily need to use resistors there, but just connect them by a thin trace/wire and that way the schematic of these stages will be unchanged electrically.
as for my transformer, the 18+18V is not the ac output but what is obtained after rectification, so it will drop even lower when there is signal. the transformer actually delivers 14+14Vac. that is not the main problem since i don't plan on going to insane levels (my room is rather small) but then i have to add the transformer for the preamp as well and it's not even the same 180V as i use now, but 300V so i have to redo the whole preamp as well and that's not particularly nice.
as a solution i will start rummaging around for any piece of mica i can use to insulate the ic's from heatsink so i'll just need to insulate the shields of the input cable from the ground by resistors as mentioned and the preamp will receive ground for power via a separate wire from the power supply. that should take care of any possible ground loop. As for the grounds in the output stages, i suppose i don't necessarily need to use resistors there, but just connect them by a thin trace/wire and that way the schematic of these stages will be unchanged electrically.
ok forget the loop nonsense, now the problem just made a sharp turn for the worse. i realised pin 5 was not connected out of my neglect. i assumed it would be internally connected but it is not. now that it is connected, the ics heat up momentarily and of course then the power supply protection trips. before it trips the voltage across the ic is only 25V (out of 60) that indicates something really bad just not sure what.
Sorry about that, must have neglected it out of hurry.
On the other hand they do work now, but depending on volume, the frequency response seems to change. I feel necessary to add that it's the volume of the sound card that i am changing while the actual potentiometer stays the same. I don't imagine there's a chance it's the amp's fault but i thought i'd just ask to be sure.
On yet another hand (maybe i'll grow one, might be useful) i need to ask, since i never could fully grasp or at least i think i still haven't grasped the issue of ground loops. So below you will see a partial (it was quicker this way) schematic of what i did in the amplifier. So at least in theory, the signal from the preamp to the IC output is not disturbed by the output signals themselves and if it needs to still reach pin 4 it can do so, again at least in theory, via the 10ohm resistor, since it "sees" it as a shorter path (low impedance for audio signals). Meanwhile, in theory, the power output signals should be confined to the power path since it has much lower resistance than 10ohm even though it is longer. But my question is, does that in fact happen or not entirely? As you can see, and i believe it's inevitable in any amp, the grounds do come together farther away, in the power supply. Also, i did add 10ohm resistors also on the input ground for the same loop reason, and also i wanted to be sure the signal would use the signal cable shields not the ground of the wall outlet (not sure which presents more of an attraction to the audio). Now i have this idea to connect the safety ground via a small coil but with thick wire as not to defeat it's purpose. This way high frequency audio that is most prone to suffer will not be allowed to travel through safety ground. So maybe you fine people can tell me if i'm doing anything right or a total mess.
Rushing things often causes mishaps: yet another misstake, the ground from before the power supply is NOT the same as the one after, so don't get alarmed.
Last edited:
This is opposite to what is needed.
A capacitor between Chassis and Audio Ground will appear as a short circuit to RF interference. This is what attenuates the RF from appearing on your Audio. It allows the chassis to operate as a screen or Faraday Cage.
Thanks for the idea. So i take it then mine is more like an upside down cake. The power supply is placed in a large cage to prevent escaping RF, which is connected to it's ground , the one in the drawing before it distributes to preamp and power ground. And since the power supply is attached to the chassis (which is an old AT computer case) then the chassis is at ground potential but... not as you (or any professional) would have done it. Although this may not be my main problem. Or maybe RF is indeed a problem and i never realised it...
edit: and the safety ground is connected (at least for the moment) to the power supply cage, essentially the power ground.
another edit: now i see the more i try to learn about important details the more stupid i feel. i used to just slap some ic's together and it would play but since i found out there's a true "art" to making everything right, i can't seem to ever be satisfied.
edit: and the safety ground is connected (at least for the moment) to the power supply cage, essentially the power ground.
another edit: now i see the more i try to learn about important details the more stupid i feel. i used to just slap some ic's together and it would play but since i found out there's a true "art" to making everything right, i can't seem to ever be satisfied.
Last edited:
Oh please, don't worry, that's the step by step learning process.
And as the old sage saying goes: "Rome was not built in one day"
I'd ask you to clear some values in your schematic, otherwise it's impossible to comment on "depending on volume, the frequency response seems to change."
The first triode shows no plate resistor , and cathode resistor value, those on the second triode, the volume control etc, aren't identified either.
Can you also post voltages ?
As of "frequency response" variation, *if* what you hear is perceived less bass and somewhat less treble at low volumes, it's a normal response of our ear.
Google "Fletcher Munson" for a better explanation.
And as the old sage saying goes: "Rome was not built in one day"
I'd ask you to clear some values in your schematic, otherwise it's impossible to comment on "depending on volume, the frequency response seems to change."
The first triode shows no plate resistor , and cathode resistor value, those on the second triode, the volume control etc, aren't identified either.
Can you also post voltages ?
As of "frequency response" variation, *if* what you hear is perceived less bass and somewhat less treble at low volumes, it's a normal response of our ear.
Google "Fletcher Munson" for a better explanation.
Nice of you to take the time to assist a beginner. Not actually a beginner but just been more careless until now.
Here's the complete (i hope) schematic of one channel:
No, it's not perceived loss of bass. Again, rushing things i did not specify exactly what happens. The problem is, stereo imaging seems a bit unstable. One minute everything seems fine, then after that first minute give or take treble seems more on the left, on the right, voice pans left, then right (mids). This happens as the particular sound pans naturally in the material played but instead of gently shifting, only slightly, as it does in the actual song, these frequencies seem to make too much of an abrupt jump, rarely are they in the middle where they should be. The bass is fine, slightly more mid bass on the left and more low bass on the right but that i can attribute to speaker placement as well as speaker properties. The speakers do a nice job but not precisely audiophile grade, more of an entry level system, Pioneer CS5070. I can also sense the woofer on the right speaker is softer, vibrates more readily than the other, so that accounts for more low bass. I have to admit, they do not rate these to go very low so as per specs they behave normally. Now stangely enough, the "tests" on youtube all seem fine except when going beyond 6KHz when treble seems to bounce around at times from left to right. thing is i do not trust these since youtube ruins the original audio quality.
About the schematic (and feel free to try it, it is fantastic, if you ignore the little problem i have) You can see these triodes operate at just under 4mA so that gives them a long lifetime, but still even so, they go well beyond human hearing range. As you can see the tone controls pots are not audio taper so i had to "fake" it with the 220K resistors. Resistors of plates and cathodes are all 1W (though only 0.5W would have done). The output capacitors on each plate are connected to it via small (2-3cm) shielded cable with the shield at HT potential. This also acts as bandwidth limiting capacitors because these triodes are a.f.a.i.k intended for RF usage. The tone circuit itself you'll probably find so easily since it's courtesy of Google image search (Harman-Kardon imitation). As they eloquently put it, there still is need for a grid leak resistor around 500K to satisfy the impedance condition of the circuit. Now to the main volume pot. The impedance on ouput of preamp is below 5K (at least it should) but that is not why the 91K resitor is there. Before the LM3886T i was using a transistor output that required a big input signal to go to full spec power and also had a reasonably large impedance at input, which the LM3886 does not quite match. Furthermore, the 91K resitor helps to overcome the big messy noise this so highly acclaimed Creative sound card has, in essence i push more signal from the sound card and in doing so the signal/noise ratio increases. At the same time, the preamp sees a higher impedance at output. Now i also had to use 4.7uF capacitors for coupling (because of apparently impedance issues - at least before i planted the 91K there) which are paper and are so large they don't actually fit on the board.
Now the amp section you know all too well. First thing i did was to eliminate the "optional components" thinking i would get rid of the awkward response issue. It only improved frequency response slightly but the response shifting remained. These IC's are not genuine as i can tell and so they have a much smaller bandwidth than the originals, they wouldn't oscillate even if i begged them. Still i decided to leave the gain as it is in the "typical application", in spite of wanting a lower one, for fear of any possible stability consequences. I don't know about you but i prefer stability at any volume at the cost of some minor hissing as the computer "thinks" or things like that. Ok, so the output is typical (only without the RL, maybe i'll add it later today). There's MKP caps to bypass the large electros and a switch that i use to couple the speakers right after power-up. The IC supresses transients but once it "starts" the electros will still charge abruptly causing a large boom, that can't be helped. I used a relay circuit before but sadly the contacts melted because of the large transient of the LM. The transistors used for the "fake ground" on pin 7 are measured to have exactly the same DC gain so the midpoint-voltage at the output shown is the same for both IC's.
* the voltages in the preamp are give or take, as a probably long time enthusiast you know tubes are not accurate devices. i also saw they are different from what they were when i was using transistor output (values on first scan). these here were the values measured after about 2h of operation with more than half of that time having signal.
Let me know if there is anything i should detail further.
Here's the complete (i hope) schematic of one channel:
No, it's not perceived loss of bass. Again, rushing things i did not specify exactly what happens. The problem is, stereo imaging seems a bit unstable. One minute everything seems fine, then after that first minute give or take treble seems more on the left, on the right, voice pans left, then right (mids). This happens as the particular sound pans naturally in the material played but instead of gently shifting, only slightly, as it does in the actual song, these frequencies seem to make too much of an abrupt jump, rarely are they in the middle where they should be. The bass is fine, slightly more mid bass on the left and more low bass on the right but that i can attribute to speaker placement as well as speaker properties. The speakers do a nice job but not precisely audiophile grade, more of an entry level system, Pioneer CS5070. I can also sense the woofer on the right speaker is softer, vibrates more readily than the other, so that accounts for more low bass. I have to admit, they do not rate these to go very low so as per specs they behave normally. Now stangely enough, the "tests" on youtube all seem fine except when going beyond 6KHz when treble seems to bounce around at times from left to right. thing is i do not trust these since youtube ruins the original audio quality.
About the schematic (and feel free to try it, it is fantastic, if you ignore the little problem i have) You can see these triodes operate at just under 4mA so that gives them a long lifetime, but still even so, they go well beyond human hearing range. As you can see the tone controls pots are not audio taper so i had to "fake" it with the 220K resistors. Resistors of plates and cathodes are all 1W (though only 0.5W would have done). The output capacitors on each plate are connected to it via small (2-3cm) shielded cable with the shield at HT potential. This also acts as bandwidth limiting capacitors because these triodes are a.f.a.i.k intended for RF usage. The tone circuit itself you'll probably find so easily since it's courtesy of Google image search (Harman-Kardon imitation). As they eloquently put it, there still is need for a grid leak resistor around 500K to satisfy the impedance condition of the circuit. Now to the main volume pot. The impedance on ouput of preamp is below 5K (at least it should) but that is not why the 91K resitor is there. Before the LM3886T i was using a transistor output that required a big input signal to go to full spec power and also had a reasonably large impedance at input, which the LM3886 does not quite match. Furthermore, the 91K resitor helps to overcome the big messy noise this so highly acclaimed Creative sound card has, in essence i push more signal from the sound card and in doing so the signal/noise ratio increases. At the same time, the preamp sees a higher impedance at output. Now i also had to use 4.7uF capacitors for coupling (because of apparently impedance issues - at least before i planted the 91K there) which are paper and are so large they don't actually fit on the board.
Now the amp section you know all too well. First thing i did was to eliminate the "optional components" thinking i would get rid of the awkward response issue. It only improved frequency response slightly but the response shifting remained. These IC's are not genuine as i can tell and so they have a much smaller bandwidth than the originals, they wouldn't oscillate even if i begged them. Still i decided to leave the gain as it is in the "typical application", in spite of wanting a lower one, for fear of any possible stability consequences. I don't know about you but i prefer stability at any volume at the cost of some minor hissing as the computer "thinks" or things like that. Ok, so the output is typical (only without the RL, maybe i'll add it later today). There's MKP caps to bypass the large electros and a switch that i use to couple the speakers right after power-up. The IC supresses transients but once it "starts" the electros will still charge abruptly causing a large boom, that can't be helped. I used a relay circuit before but sadly the contacts melted because of the large transient of the LM. The transistors used for the "fake ground" on pin 7 are measured to have exactly the same DC gain so the midpoint-voltage at the output shown is the same for both IC's.
* the voltages in the preamp are give or take, as a probably long time enthusiast you know tubes are not accurate devices. i also saw they are different from what they were when i was using transistor output (values on first scan). these here were the values measured after about 2h of operation with more than half of that time having signal.
Let me know if there is anything i should detail further.
figured out part of the problem: volume pot cheap chinese garbage making the volume "slip" to maximum when pressed slightly towards the left. discovered this accidentally, almost blew a speaker i think. although i can't see why that would affect frequency response. i suppose poor contact of the wiper or one of the ends would manifest in noise and/or prevent some frequencies getting through... so are the tone control ones but they seem to work without "running amok" i'm not touching them ever again. i ordered some Alphastat pots from Conrad but it will be a while until they arrive also decided to add balance control i had also eliminated due to instability of a pot.
edit: also eliminated high pitched noise from sound card - it was not actually from the sound card but as i suspected the audio had a shorter path (or of smaller resistance) via the outlet ground and was picking up all the noise from all the switching power supplies in that outlet including that of the amp itself. added low pass coil of 2mm gage wire i had made for a crossover filter. it prohibits high bass mids and highs from travelling thru there while safety ground remains in effect (there is no tingling when touching the amp case).
makes sense that it would be the pot to blame, it seems the problem slightly diminishes if i push more signal on the input, or perhaps since the music is louder i can't really tell the difference that much... wish i had some actual instruments to measure output signal
also decided to add balance control i had also eliminated due to instability of a pot.edit: also eliminated high pitched noise from sound card - it was not actually from the sound card but as i suspected the audio had a shorter path (or of smaller resistance) via the outlet ground and was picking up all the noise from all the switching power supplies in that outlet including that of the amp itself. added low pass coil of 2mm gage wire i had made for a crossover filter. it prohibits high bass mids and highs from travelling thru there while safety ground remains in effect (there is no tingling when touching the amp case).
makes sense that it would be the pot to blame, it seems the problem slightly diminishes if i push more signal on the input, or perhaps since the music is louder i can't really tell the difference that much... wish i had some actual instruments to measure output signal
Last edited:
Well, any real cheap multimeter , those with only 2 AC scales, 200VAC and 750VAC will measure speaker output, and a somewhat better one (with a 2VAC or even better, 200mVAC scale) can measure preamp level signals.
As of a sinewave generator, there are very simple circuits if you dare build an oscillator but you can download free software to turn your PC into a generator.
Check the *IMPRESSIVE* work of Chris Bonis a.k.a. "xstreme" : http://www.diyaudio.com/forums/software-tools/225367-tgen-v1-0-free-software-function-generator.html
Read the full thread because he upgraded it many times adding special functions "on demand".
I'd start with the basic one and in the future upgrade, if needed.
If you don't have a computer close by, download a couple MP3 sinewave tones, at least 1KHz and 400 or 440Hz , the most used ones., and play it with a portable MP3 player.
Most can supply around 200mV RMS .
So you can inject a predictable signal at the input and take a useful measurement at the output.
As of a sinewave generator, there are very simple circuits if you dare build an oscillator but you can download free software to turn your PC into a generator.
Check the *IMPRESSIVE* work of Chris Bonis a.k.a. "xstreme" : http://www.diyaudio.com/forums/software-tools/225367-tgen-v1-0-free-software-function-generator.html
Read the full thread because he upgraded it many times adding special functions "on demand".
I'd start with the basic one and in the future upgrade, if needed.
If you don't have a computer close by, download a couple MP3 sinewave tones, at least 1KHz and 400 or 440Hz , the most used ones., and play it with a portable MP3 player.
Most can supply around 200mV RMS .
So you can inject a predictable signal at the input and take a useful measurement at the output.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.